Hugging Face's logo

appmana
/
deepseek-v4-mxfp4-int8

Safetensors
deepseek_v4
8-bit precision
dsv4_mxfp4_int8
Model card Files
xet
 Community
doctorpangloss commited on
Commit
d62259e
·
verified ·
1 Parent(s): 8bff79a

Regenerate: fused layout (1593 tensors, 31 shards), mxfp4 experts + int8 channel dense, appmana kernel block

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. config.json +15 -13
  2. model-00001-of-00072.safetensors → model-00001-of-00031.safetensors +2 -2
  3. model-00002-of-00072.safetensors → model-00002-of-00031.safetensors +2 -2
  4. model-00003-of-00072.safetensors → model-00003-of-00031.safetensors +2 -2
  5. model-00005-of-00072.safetensors → model-00004-of-00031.safetensors +2 -2
  6. model-00005-of-00031.safetensors +3 -0
  7. model-00006-of-00031.safetensors +3 -0
  8. model-00006-of-00072.safetensors +0 -3
  9. model-00007-of-00031.safetensors +3 -0
  10. model-00008-of-00031.safetensors +3 -0
  11. model-00008-of-00072.safetensors +0 -3
  12. model-00009-of-00031.safetensors +3 -0
  13. model-00010-of-00031.safetensors +3 -0
  14. model-00010-of-00072.safetensors +0 -3
  15. model-00011-of-00031.safetensors +3 -0
  16. model-00011-of-00072.safetensors +0 -3
  17. model-00012-of-00031.safetensors +3 -0
  18. model-00013-of-00031.safetensors +3 -0
  19. model-00013-of-00072.safetensors +0 -3
  20. model-00014-of-00031.safetensors +3 -0
  21. model-00015-of-00031.safetensors +3 -0
  22. model-00015-of-00072.safetensors +0 -3
  23. model-00016-of-00031.safetensors +3 -0
  24. model-00016-of-00072.safetensors +0 -3
  25. model-00017-of-00031.safetensors +3 -0
  26. model-00018-of-00031.safetensors +3 -0
  27. model-00018-of-00072.safetensors +0 -3
  28. model-00019-of-00031.safetensors +3 -0
  29. model-00019-of-00072.safetensors +0 -3
  30. model-00020-of-00031.safetensors +3 -0
  31. model-00021-of-00031.safetensors +3 -0
  32. model-00021-of-00072.safetensors +0 -3
  33. model-00022-of-00031.safetensors +3 -0
  34. model-00023-of-00031.safetensors +3 -0
  35. model-00023-of-00072.safetensors +0 -3
  36. model-00024-of-00031.safetensors +3 -0
  37. model-00024-of-00072.safetensors +0 -3
  38. model-00025-of-00031.safetensors +3 -0
  39. model-00026-of-00031.safetensors +3 -0
  40. model-00026-of-00072.safetensors +0 -3
  41. model-00027-of-00031.safetensors +3 -0
  42. model-00028-of-00031.safetensors +3 -0
  43. model-00028-of-00072.safetensors +0 -3
  44. model-00029-of-00031.safetensors +3 -0
  45. model-00029-of-00072.safetensors +0 -3
  46. model-00030-of-00031.safetensors +3 -0
  47. model-00031-of-00031.safetensors +3 -0
  48. model-00031-of-00072.safetensors +0 -3
  49. model-00032-of-00072.safetensors +0 -3
  50. model-00034-of-00072.safetensors +0 -3
config.json CHANGED
@@ -53,9 +53,8 @@
53
  "type": "float"
54
  },
55
  "targets": [
56
- "*.ffn.experts.*.w1",
57
- "*.ffn.experts.*.w2",
58
- "*.ffn.experts.*.w3"
59
  ]
60
  },
61
  "linears_w8a16": {
@@ -70,19 +69,14 @@
70
  "type": "float"
71
  },
72
  "targets": [
73
- "*.attn.wq_a",
74
  "*.attn.wq_b",
75
- "*.attn.wkv",
76
  "*.attn.wo_a",
77
  "*.attn.wo_b",
78
  "*.attn.indexer.wq_b",
79
  "*.attn.indexer.compressor.wkv",
80
- "*.attn.indexer.compressor.wgate",
81
- "*.ffn.shared_experts.w1",
82
- "*.ffn.shared_experts.w2",
83
- "*.ffn.shared_experts.w3",
84
- "mtp.*.e_proj",
85
- "mtp.*.h_proj"
86
  ]
87
  }
88
  },
@@ -164,5 +158,13 @@
164
  128,
165
  4,
166
  0
167
- ]
168
- }
 
 
 
 
 
 
 
 
 
53
  "type": "float"
54
  },
55
  "targets": [
56
+ "*.ffn.experts.routed_experts.w13_weight",
57
+ "*.ffn.experts.routed_experts.w2_weight"
 
58
  ]
59
  },
60
  "linears_w8a16": {
 
69
  "type": "float"
70
  },
71
  "targets": [
72
+ "*.attn.fused_wqa_wkv",
73
  "*.attn.wq_b",
 
74
  "*.attn.wo_a",
75
  "*.attn.wo_b",
76
  "*.attn.indexer.wq_b",
77
  "*.attn.indexer.compressor.wkv",
78
+ "*.ffn.shared_experts.gate_up_proj",
79
+ "*.ffn.shared_experts.down_proj"
 
 
 
 
80
  ]
81
  }
82
  },
 
158
  128,
159
  4,
160
  0
161
+ ],
162
+ "appmana": {
163
+ "kernels": [
164
+ "flash_mla.flash_sparse_mla_decode",
165
+ "flash_mla.triton_sparse_int8_mla_decode",
166
+ "vllm.models.deepseek_v4.nvidia_sm86.triton_kernels.sparse_attention_triton"
167
+ ],
168
+ "cache_type": "fp8_ds_mla"
169
+ }
170
+ }
model-00001-of-00072.safetensors → model-00001-of-00031.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5176586613905d4beaadbaea1cfecd1693e17f9da4a0ea09d99aab7d3f2f5b7c
3
- size 1059061856
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dfa060ed52e9b88a149c508c306a1a43536abceac26264f35b29d9051e14adf3
3
+ size 5601513832
model-00002-of-00072.safetensors → model-00002-of-00031.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2ac59cf4b450cf942a8eafca40716268836beaaca6b77ef3a763e53df54592bc
3
- size 3566422568
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0824bb558fb1232872733c73f26ac5f8cd0b6ecd03acd5839b82ae47d4116a8e
3
+ size 4661989872
model-00003-of-00072.safetensors → model-00003-of-00031.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2ed22bad27499d4d33f89360ba8eaacbde96d0baaba39cefbdcba4afa1acfaa3
3
- size 3566422568
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0cfad1e5c0e9c2a2b3bcf38254ad7612404cc718e73614d16d449c8d27f4afb
3
+ size 5700100264
model-00005-of-00072.safetensors → model-00004-of-00031.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f2358ab248589b043282dcd5ba1b6e33f2da86cafaa59011d6357b4d228e3da2
3
- size 3596346520
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5c9da066dfb072e9366c72d528e2125d1909b4a0c1949045dfda2ff71b3014a4
3
+ size 4661989872
model-00005-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e0923a72136683519f54c0176395448d46f840d0e8259077e67c93b3f396e41a
3
+ size 5700100264
model-00006-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e548b7483aa2c1a012c796ba2810d203484acf4273e6452731fe5a4b4f6daf7
3
+ size 4661989872
model-00006-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:61eab6378abf5f67a84dc0dc6d999916d56226655f44544ea9eeca2e4f69b709
3
- size 3568870360
 
 
 
 
model-00007-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97a1b0898671b2af9f48ea735d7207557ad9a8aac8aef7a597c0cfdd47dbedb5
3
+ size 5700100264
model-00008-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40e3c666a80813e9634e4bbeb57fd24e1ecf19bc967123e3a28c250a94e2fbcb
3
+ size 4661989880
model-00008-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:00299b79b5fb3f2b75a18770b7503daa2f43a065359e08f190b03f4cde2170ef
3
- size 3590142024
 
 
 
 
model-00009-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9edf9a0a0a2e8cfd17af12cbf81bf83766648a2f4e062d621d85d0bff93b0479
3
+ size 5700100280
model-00010-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61a5daf2c7f2ebe162d5aa4731d5de582ca0f5229bfd9c5473d9bfd2de12fc65
3
+ size 4661989880
model-00010-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:4d9851d0334df9242bdc7b1849f49be37cb041b4b6f0a98423590df3124515a0
3
- size 3568870360
 
 
 
 
model-00011-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c53cff0d496e1d674f94fadec75e703f6bbfd3be74288dc19858bb50181d5f7b
3
+ size 5700100280
model-00011-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:5069debb8837c27a4a8b241c7469087e2b206bebab17b0d03c5847b2053c5915
3
- size 3590142024
 
 
 
 
model-00012-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb5e2b49080204a1203a55f0c92a65eb9b184e711ee0a088d93a4bf6e1f3d7da
3
+ size 4661989880
model-00013-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b289f8c85fa9029a5aa8b636e332838073c0c50ce3815da5c6f3f92612d2d870
3
+ size 5700100280
model-00013-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:ecb69faca37f98d4c2b24b4f163ca8bc0df68b1c12122074c1f2b10ac298048d
3
- size 3568870360
 
 
 
 
model-00014-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59c3985e3ab68288a3244c60ce927ba2eeba0a6a4bfd64e8e779f71c45c7dab4
3
+ size 4661989880
model-00015-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8cd2836df2a273ad8b1aee57c39cb6f8712d4fc662c6084fae359e0150e22b1d
3
+ size 5700100280
model-00015-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b17455f23fb1e4871feb473ae7ae3e36a3051785bc22422cc86b5e24a1818e25
3
- size 3590142024
 
 
 
 
model-00016-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d13974b70e346067a3cd51ac6a96acd16bca15832b2b5dda265d0dd08a0c0330
3
+ size 4661989880
model-00016-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:07b2d9b358e6b504e162426bbd8865704a4a7864c96929eb1c8743feed3ee3e1
3
- size 3568870360
 
 
 
 
model-00017-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c2f72b9429dab022841c9df3d98e70b51240253b370e0e174f9487c441808b1
3
+ size 5700100280
model-00018-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:75d7958ae56ca983e7feef31b81b90430909742184d15c58bf937b40a1962b60
3
+ size 4661989880
model-00018-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0752efb794d5e91a9ee169e528419534c1ae689fd408649575ee018bdedb01a9
3
- size 3590143600
 
 
 
 
model-00019-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99ad90cbece9fceaa2812b04b8d7edd2484c957a8d117368ce93b2717ef1b996
3
+ size 5700100280
model-00019-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f2a12189a0c2af660021c8d01076e7b05d10ef454a7db05c59582ebba600b6a7
3
- size 3568871928
 
 
 
 
model-00020-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af3e0ba51fb88185f68a29435714797021d9be3bde46ca5d69bffb1183c768d7
3
+ size 4661989880
model-00021-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ecfa5b33e04108343e89cd45267988498174ee7495fc265c2bd028c19af0fb1
3
+ size 5700100280
model-00021-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:d88b47510a18dfd34f2a5c0cb2164a36d790cce9bee946e03b418e8532415570
3
- size 3590143600
 
 
 
 
model-00022-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:32c1f1383a107bfd01f68bf5da319ca41c4db874d96e2e6bb2b76ff7f24c1123
3
+ size 4661989880
model-00023-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c30922ff2b4a81666afb03216be7133e786891e1be18143694c4e1cd564d134f
3
+ size 5700100280
model-00023-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:56962f7bcd617156d5c095620798ec44a2b84e623c70c7a6451d5a1d8db50dad
3
- size 3568871928
 
 
 
 
model-00024-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40a72f60ac369eff5704ae24f6efcfd5bba4b1567d21cb0d7b80d8f792914d4c
3
+ size 4661989880
model-00024-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:53297e441b982c15887f64770efc1cb94191473952250d11c6f41c66fd75e094
3
- size 3590143600
 
 
 
 
model-00025-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d373293ffb5ea2103e07369df8e800cd9110852cdeab8b0483cfecb5a318d057
3
+ size 5700100280
model-00026-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97ede80785e7576f33c80a804dbf41d4e85b6a68fcfe7948987d096a2ee6b578
3
+ size 4661989880
model-00026-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:9bd1220d326945ccb70047dfcb3c3e2b2dc0d11dee88da1e02fb71592859ec7a
3
- size 3568871928
 
 
 
 
model-00027-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4697ce2d1427f97efa43633407fc8fc7fa59fd71d6993001b90fdf34d10c1adf
3
+ size 5700100280
model-00028-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:903504c24e47a3b14d86d26b48dd68c783cebd9eb56dc93b119f86f9c7a120d1
3
+ size 4661989880
model-00028-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:eff35d7fd2266939672d0373bf37d9fadaa6e005e7159f16928fbb544a5e3e68
3
- size 3590143600
 
 
 
 
model-00029-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9c19b58ef3608ffd2cbb7790c12798d9bd0e57ebec04983467b54f5fc656dc1
3
+ size 4611678432
model-00029-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3d05ce0fe3e535f8fe683793fb6df8debcab59c2e097231fc7913d34896e63a6
3
- size 3568871928
 
 
 
 
model-00030-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3498550e375f08ba00d48372e8501523cd1c011141338cf71b452fe7ca8ddf13
3
+ size 4295773404
model-00031-of-00031.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d8d86cc176b94796b41833294e4255e21a5f5cc42b2b8670281ef84616133a1f
3
+ size 2282315300
model-00031-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:5016bfce7afba954f0e1411422c06dae6c07e9166b0e2b1bae0736cc62d6108d
3
- size 3590143600
 
 
 
 
model-00032-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:bb24cd23fd771794efd00f2a85de6140d46b6b754ca88d9be94f726da429cd11
3
- size 3568871928
 
 
 
 
model-00034-of-00072.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0e2e4a105d45d803287ddfb38bad7ab44225824cc7e2806dc0df2ea6fd4de1c9
3
- size 3590143600